Powernap should ensure WoL is enabled before Second Stage

Thank you for reporting this bug. Can you please run "apport-collect 989175" to collect some information on your system so that it might be easier to diagnose? Also, what does "WOL" mean?

apport information

apport information

apport information

apport information

Added the apport information.

WOL=WakeOnLAN

thanks.

Body:

Thanks. I was just making sure ;)

 summary "Powernap prevents waking from WakeOnLAN"

Sincerely,
kroq-gar78

OK, I think I have found the reason for this bug. It is not due to stage 2 in power nap, but rather stage 1 - during powersave mode the ethernet configuration for the card is changed, which deactivates wake-on-lan.

Since power nap first enters in stage1 then stage2, when the computer is turned off (or suspended) in stage 2, the wake-on-lan is deactivated... Of course I do *not* see this when manually shutting down, because my typing on the console or through ssh gets the computer out of powersave mode.

Here is the log file which demonstrates this. It is produced by a while loop which logs the end of /var/log/powernap.log and the result of "ethtool eth0" every 5 seconds. As you can see below the first two logged entries (at t=0s and t=5s) have "Wake-on: g" but at t=+10s there is "Wake-on: d" which I assume means deactivated...

=> I am not sure where power nap changes this, isn't it supposed to just change speed ?

####### Time series of /var/log/powernap.log and "ethtool eth0" at 5s intervals #########

samedi 28 avril 2012, 11:24:08 (UTC+0200)
2012-04-28_11:23:45 WARNING Entered into GRACE PERIOD. Action [/usr/sbin/powernap] will be taken in [3] seconds
2012-04-28_11:23:46 WARNING Entered into GRACE PERIOD. Action [/usr/sbin/powernap] will be taken in [2] seconds
2012-04-28_11:23:46 WARNING Taking action [/usr/sbin/powernap]
2012-04-28_11:23:50 WARNING Taking recover action [/usr/sbin/pm-powersave false]
2012-04-28_11:23:59 WARNING Entered into GRACE PERIOD. Action [/usr/sbin/powernap] will be taken in [6] seconds
2012-04-28_11:24:00 WARNING Entered into GRACE PERIOD. Action [/usr/sbin/powernap] will be taken in [5] seconds
2012-04-28_11:24:02 WARNING Entered into GRACE PERIOD. Action [/usr/sbin/powernap] will be taken in [4] seconds
2012-04-28_11:24:05 WARNING Entered into GRACE PERIOD. Action [/usr/sbin/powernap] will be taken in [6] seconds
2012-04-28_11:24:06 WARNING Entered into GRACE PERIOD. Action [/usr/sbin/powernap] will be taken in [5] seconds
2012-04-28_11:24:08 WARNING Entered into GRACE PERIOD. Action [/usr/sbin/powernap] will be taken in [4] seconds
Settings for eth0:
 Supported ports: [ TP MII ]
 Supported link modes: 10baseT/Half 10baseT/Full
                         100baseT/Half 100baseT/Full
                         1000baseT/Half 1000baseT/Full
 Supported pause frame use: No
 Supports auto-negotiation: Yes
 Advertised link modes: 10baseT/Half 10baseT/Full
                         100baseT/Half 100baseT/Full
                         1000baseT/Half 1000baseT/Full
 Advertised pause frame use: Symmetric Receive-only
 Advertised auto-negotiation: Yes
 Link partner advertised link modes: 10baseT/Half 10baseT/Full
                                      100baseT/Half 100baseT/Full
                                      1000baseT/Half 1000baseT/Full
 Link partner advertised pause frame use: Symmetric Receive-only
 Link partner advertised auto-negotiation: Yes
 Speed: 1000Mb/s
 Duplex: Full
 Port: MII
 PHYAD: 0
 Transceiver: internal
 Auto-negotiation: on
 Supports Wake-on: pumbg
 Wake-on: g
 Current message level: 0x00000033 (51)
          drv probe ifdown ifup
 Link detected: yes

samedi 28 avril 2012, 11...

Hi there,

Thank you for taking the time to look into this. This is actually not a PowerNap bug, but rather, pm-utils (not really a bug though). pm-utils ships a disable_wol script that, as its name says, disables wake-on-lan when executing the pm-powersave (which powernap executes).

To fix this, you could either disable this script when pm-powersave executes by removing the execution bit form the script. You'll find the script at:

/usr/lib/pm-utils/power.d/disable_wol

Now, what should be done in PowerNap though, is to ensure WoL is enabled when performing a poweroff/suspend action. I'll retarget this bug for that.

Cheers.

Thanks for the quick answer ! Indeed it is an issue between powernap and pm-utils. I had looked in /etc/pm but did not think to look in /usr/lib/pm-utils.

I really think this disable_wol should be deactivated by default (how much power does that save ?) - many people will want to use auto suspend/poweroff with WoL enabled for servers.

And by the way, thanks a lot for powernap, it is really useful ! Can't wait for the autowake-up using ARP :-)

So should I set it as affecting 'pm-utils' too?

@kroq-gar78: I think it would be better to file a new bug.

@vincefn: The "PowerNap Server" features (That includes the ARP monitors to autowake PoweNap clients) have been disabled in Precise because it was only a preview of what could be done with PowerNap and I didn't have the time to finish it. I however, can re-enable it and make it available in PPA for you to use.

@Andres - thanks, it would be nice to be able too test the wake-on-ARP, so putting this in a PPA looks like a great idea.

This being said, I just realized that my server does not have the "Wake-On: a" support that is required, so I won't be able to test this unless I change the card :-(

I agree with comment #11 - is disable_wol really worth the trouble? does it really save energy? if not, let's disable the script.